Abstract:
The IT industry is rapidly developing due to technological advancements, and backend
and frontend developers are playing a crucial role in its success. However, high turnover
rates among these developers are a significant challenge for companies, and identifying
the factors that influence employee turnover is crucial. A study was conducted
using an online survey questionnaire and 203 data were collected. Logistic Regression
was used to assess the significant variables which affect the turnover intentions and
a machine learning model was developed to predict turnover intentions. The process
involved collecting data through an online questionnaire, pre-processing and feature
engineering, splitting the data into training and testing sets, building and evaluating
various statistical machine learning algorithms using accuracy score, precision, recall
and confusion matrix, selecting the best model, saving it as a pickle file, and hosting it
using AWS Lambda, AWS S3, and AWS API GATEWAY services. A web interface was
also created to provide users with analysis and feature importance graphs as well as the
ability to make predictions by uploading evaluation sheets. Predictions were captured
through an API call to the hosted model, and the results could be downloaded in a .csv
format. The results showed that job satisfaction, recognition, organizational support,
and organizational commitment had a moderate negative relationship with turnover intention.
In contrast, alternative job opportunities and job stress had a moderate positive
relationship. Organizational commitment, job stress, and alternative job opportunities
were identified as the most significant variables affecting turnover intentions. Support
Vector Machine was identified as the best model with 100% test accuracy and it was
hosted in cloud services to get the predictions. Overall, this approach could be useful in
identifying employees at risk of turnover and in enabling companies to take proactive
measures to retain their valuable employees.
